The Agent Skills Directory

[PROMPT_INJECTION]: No direct prompt injection patterns (e.g., DAN, instruction overrides) were found. The skill has a low-risk surface for indirect prompt injection as it processes external data.
Ingestion points: Reads docs/PRD.md to understand product goals.
Boundary markers: Uses structured Markdown headers to separate analysis steps, though it lacks explicit instructions to ignore embedded commands in the data.
Capability inventory: Limited to file reading and generating text-based reports; no network or execution capabilities are utilized.
Sanitization: Not specified in the instructions.
[DATA_EXFILTRATION]: The skill accesses local project paths (docs/PRD.md, app/, components/) to perform analysis. No network operations or patterns indicating the transmission of this data to external servers were detected.
[REMOTE_CODE_EXECUTION]: There are no references to downloading or executing external scripts, nor any dynamic code generation patterns.
[COMMAND_EXECUTION]: The skill defines a logical workflow for the agent to follow but does not include instructions to execute system-level commands or shell scripts.
[CREDENTIALS_UNSAFE]: No hardcoded credentials, API keys, or attempts to access sensitive system files (like SSH keys or environment variables) were found.

ux-walkthrough-skill